High Serum PCSK9 Is Associated With Increased Risk of New-Onset Diabetes After Transplantation in Renal Transplant Recipients

In renal transplant recipients (RTRs), new-onset diabetes after transplantation (NODAT) is a major metabolic complication (1). Its incidence has increased during the past decades (1,2). Estimates of the frequency of NODAT vary widely and have been reported to range from 2% to 50% (3). Twelve months cumulative incidence of NODAT was highest in the era of high-dose steroid regimens in the 1970s, lowest with cyclosporine-based regimens, and again higher with tacrolimus-based regimens (3,4). Although the observation period of many studies does not go beyond an initial surge in the incidence of NODAT during the first year after transplantation, it is well documented that during the era of cyclosporine-based regimens, the largest number of incident cases of NODAT occurred beyond the first year after transplantation (1). With current tacrolimus-based regimens, the number of the incident cases that occur beyond the first year after transplantation is even higher (4).

That NODAT adversely affects patient and graft survival in RTRs is well appreciated (5,6). Therefore, better identifying which RTRs are at increased risk of developing NODAT is clinically relevant. RTRs also are prone to the development of atherosclerotic cardiovascular disease (CVD), which represents a major cause of graft loss and a leading cause of death (7). Apart from affecting CVD risk, dyslipidemia has been found to be independently associated with reduced graft survival (8). Abnormalities in cholesterol metabolism thus can be considered to play a key role in the development of CVD and graft failure in RTRs.

A meta-analysis revealed that statin therapy increases the risk of new-onset diabetes in nontransplant populations (9). Statins lower LDL cholesterol by inhibiting hydroxymethylglutaryl-CoA reductase (10), thereby increasing LDL receptor (LDLR) expression and promoting cellular cholesterol delivery (11). Conversely, the risk of type 2 diabetes is markedly reduced in patients with familial hypercholesterolemia as a consequence of LDLR mutations (12). Moreover, diabetes incidence may be linked to genetic variation in hydroxymethylglutaryl-CoA reductase, indicating a role of cholesterol metabolism in diabetes development (9).

The proprotein convertase subtilisin/kexin type 9 (PCSK9) pathway has been identified to be crucially involved in regulating hepatic LDLR expression in humans (13). PCSK9 is a secreted protease that is predominately expressed by the liver and intestine (14). PCSK9 is able to target the LDLR toward intracellular degradation, thereby preventing LDLR from recycling to the cell surface (13). Higher circulating PCSK9 levels confer a lower rate of LDL apoB catabolism (15), supporting the notion that variations in circulating PCSK9 levels are pathophysiologically relevant.

To date, the association of serum levels of PCSK9 with NODAT development has not been evaluated. Because the PCSK9 pathway plays a central role in cholesterol metabolism, this study was primarily initiated to prospectively investigate the association between serum PCSK9 levels and NODAT development in RTRs. Secondary end points assessed were the association of serum PCSK9 with all-cause mortality, cardiovascular mortality, and renal graft failure in RTRs.

In this prospective cohort study, all adult RTRs who survived with a functioning allograft beyond the first year after transplantation were considered eligible to participate. Baseline data, obtained at least 1 year after transplantation, were collected between August 2001 and July 2003 at a median of 6.0 (interquartile range [IQR] 2–11) years after transplantation. Patients with known or apparent systemic illnesses (i.e., malignancies, opportunistic infections) were excluded from participation. Six hundred six of 847 (72%) eligible RTRs gave informed consent. We excluded 105 RTRs with existing diabetes (defined as fasting plasma glucose ≥126 mg/dL [7.0 mmol/L] and/or use of glucose-lowering drugs) (6), 44 because of unavailability of samples for measurement of PCSK9, and 4 because of being a recipient of a combined kidney/liver transplant, resulting in 453 RTRs eligible for analyses. None of these 453 RTRs was a recipient of a combined kidney/pancreas transplant. Written informed consent was obtained from all participants, and the institutional review board of the University of Groningen (Groningen, the Netherlands) approved the study (METc 01/039). Relevant donor, recipient, and transplant characteristics were extracted from the Groningen Renal Transplant Database, which has been described in detail elsewhere (16). The primary end point of this study was NODAT. Secondary end points were all-cause mortality, cardiovascular mortality, and renal graft failure. Follow-up was performed for a median of 9.6 (IQR 7.1–10.2) years until April 2012. There was no loss to follow-up for the primary and secondary end points. Collection of these data are ensured by the continuous surveillance system of the outpatient clinic of our university hospital and close collaboration with affiliated hospitals.

Blood pressure was measured as the average of three automated (Omron M4; Omron Europe B.V., Hoofddorp, the Netherlands) measurements with 1-min intervals after a 6-min rest in supine. BMI was calculated as weight in kilograms divided by height in meters squared. Waist circumference was measured on bare skin midway between the iliac crest and the 10th rib.

NODAT was defined according to International Expert Panel recommendations that were based on 2003 American Diabetes Association criteria (2) (i.e., at the end of the time frame of inclusion). The diagnosis of NODAT was based on one of the following criteria: symptoms of diabetes (classic symptoms, including polyuria, polydipsia, and unexplained weight loss) plus nonfasting plasma glucose concentration ≥200 mg/dL (11.1 mmol/L), fasting plasma glucose ≥126 mg/dL (7.0 mmol/L), or use of glucose-lowering medication. Fasting was defined as no caloric intake for at least 8 h. If patients transplanted in our center are >1 year posttransplantation, as is the case for RTRs included in this study, routine follow-up consists of three monthly visits to the outpatient clinic, with check-ups including assessment of renal function and fasting or random glucose. If glucose was elevated, a confirmatory laboratory test of venous plasma was performed on a subsequent day or at the next visit, after which the diagnosis of NODAT was made. Apart from NODAT, we categorized RTRs as having impaired fasting glucose if fasting plasma glucose was ≥100 mg/dL (5.6 mmol/L) and <126 mg/dL (7.0 mmol/L) (17). The secondary end points of the study were all-cause mortality, cardiovascular mortality, and death-censored graft failure. Cause of death was obtained by linking the number of the death certificate to the primary cause of death as coded by a physician from the Central Bureau of Statistics according to the ICD-9 (18). Cardiovascular death was defined as the principal cause of death being cardiovascular in nature (ICD-9 codes 410–447). Graft failure was defined as restart of dialysis or retransplantation. Current medication was extracted from the medical records. Information about change of immunosuppressive medication over time in our transplant center is described in the Supplementary Data. Cumulative dose of prednisolone was calculated by multiplying time after transplantation by prednisolone dose at time of inclusion in the study and addition of the dose of prednisolone or methylprednisolone required for treatment of acute rejection (a conversion factor of 1.25 was used to convert methylprednisolone to prednisolone dose).

Venous blood was drawn in the morning after an 8–12-h overnight fasting period to determine serum creatinine and plasma glucose concentrations. Serum creatinine was determined by using the Jaffe reaction (MEGA AU510; Merck Diagnostica, Darmstadt, Germany). Plasma glucose was measured by the glucose oxidase method (YSI 2300 Stat Plus; Yellow Springs Instruments, Yellow Springs, OH). Estimated glomerular filtration rate (eGFR) was calculated by applying the Chronic Kidney Disease Epidemiology Collaboration equation (19). Proteinuria was defined as ≥0.5 g protein/24-h urine. Total cholesterol was determined by using the cholesterol oxidase-phenol aminophenazone method (MEGA AU510). LDL cholesterol was calculated by using the Friedewald equation (20). HDL cholesterol was measured with the cholesterol oxidase-phenol aminophenazone method on a Technicon RA-1000 (Bayer Diagnostics, Mijdrecht, the Netherlands). Plasma triglycerides were determined with the glycerol-3-phosphate oxidase-oxidase method (YSI 2300 Stat Plus). Fasting insulin levels at baseline were determined for research purposes by using an AxSYM autoanalyzer (Abbott Diagnostics). Insulin resistance was calculated by using the HOMA for insulin resistance (HOMA-IR) equation: glucose (mmol/L) × insulin (mU/L) / 22.5 (21).

Samples for measurement of PCSK9 were collected at the moment of inclusion in the study, at the same time that baseline characteristics were assessed. The samples for measurement of PCSK9 were kept frozen at −80°C until assessment. Serum PCSK9 was assessed by PCSK9 dual monoclonal antibody sandwich ELISA, with minor modifications (22) (Supplementary Data).

Data were analyzed with SPSS version 22.0 software (IBM Corporation, Chicago, IL). In all analyses, a two-sided P < 0.05 was considered significant. Data are expressed as mean ± SD for normally distributed variables and as median (IQR) for variables with a skewed distribution. Linear regression analysis was performed to determine whether patient characteristics varied across tertiles of PCSK9 by calculating P values for trend. Residuals were checked for normality and log-transformed when appropriate. χ² test was used for categorical variables. NODAT development was visualized by Kaplan-Meier curves according to the highest versus the two lowest tertiles of PCSK9 levels, with statistical significance tested by log-rank (Mantel-Cox) test.

To study whether PCSK9 was associated with risk of NODAT, all-cause mortality, cardiovascular mortality, and renal graft failure, Cox proportional hazards regression analyses were performed. We first performed crude analyses (model 1) followed by adjustment for age and sex (model 2) and further cumulative adjustment for eGFR, proteinuria, and time since transplantation (model 3). Because Cox regression models should not contain more than one independent variable for every 10 outcome events (23), model 3 was further adjusted for potential confounders in additional Cox regression models. Adjustment variables were selected on the basis of the literature (age, sex, eGFR, proteinuria, and time since transplantation), a significant association with PCSK9 (Table 1), or a significant association with development of NODAT (Supplementary Table 1). If closely related variables within a biological domain (e.g., BMI and waist circumference in the domain of body composition) were associated with PCSK9 and/or NODAT, the variable with the strongest association or the biologically most relevant variable was primarily adjusted for to prevent occurrence of collinearity and overfitting of models. In primary analyses, therefore, we additionally adjusted for total cholesterol, triglycerides, and BMI in model 4; systolic blood pressure and fasting plasma glucose in model 5; smoking and alcohol use in model 6; statin use in model 7; HOMA-IR in model 8; and use of tacrolimus, use of cyclosporine, and trough levels of both in model 9. In the Cox regression analyses, serum PCSK9 was used both as a continuous variable and as a categorical variable in which tertiles 1 and 2 were combined and compared with tertile 3. NODAT and graft failure were censored at the date of the last follow-up or death. As secondary analyses, we repeated the Cox regression analyses for the association of serum PCSK9 with NODAT, with adjustment for the variables in biological domains that were associated with PCSK9 or development of NODAT but that were not adjusted for in the primary analyses (Supplementary Table 2).

To allow for comparison with studies in which time-to-event data were not available, we proceeded with sensitivity analyses in which we applied logistic regression rather than Cox regression. We also performed sensitivity analyses in which we investigated potential effect modification of the association of PCSK9 with NODAT by use of statins.

We included 453 stable RTRs. Mean age was 51 ± 12 years, 56% were male, and BMI averaged 25.8 ± 4.2 kg/m². Patients were included at 6.1 (2.7–11.7) years after transplantation. The mean serum PCSK9 concentration was 107.1 ± 43.4 ng/mL. Baseline characteristics of the RTRs according to tertiles of PCSK9 are shown in Table 1. Plasma PCSK9 levels were positively associated with age, BMI, waist circumference, plasma glucose, plasma insulin, HOMA-IR, serum triglycerides, and statin use. Serum PCSK9 was associated with neither total cholesterol nor LDL cholesterol and eGFR. Serum PCSK9 concentrations in RTRs treated with statins were 119.3 ± 45.7 versus 95.6 ± 37.7 ng/mL in RTRs not treated with statins (P < 0.001). Twenty nine (6%) RTRs had impaired fasting glucose at baseline.

During a median follow-up of 9.6 (7.1–10.2) years, 70 (15.5%) RTRs developed NODAT. NODAT developed in 23% (n = 35) in the upper tertile of PCSK9 versus 12% (n = 35) in the lowest two tertiles (P < 0.001 by log-rank test). The corresponding Kaplan-Meier curves are shown in Fig. 1. In Cox regression analysis, serum PCSK9 as a continuous variable was univariately associated with development of NODAT (hazard ratio [HR] 1.34 [95% CI 1.10–1.63]; P = 0.004) (Table 2). After adjustment for relevant covariates (model 2: age and sex; model 3: age, sex, eGFR, proteinuria, and time since transplantation [Fig. 2]; and subsequently added to model 3, model 4: serum triglycerides, total cholesterol, and BMI; model 5: systolic blood pressure and fasting plasma glucose; model 6: smoking and alcohol use; model 7: statin use; model 8: HOMA-IR; and model 9: use of tacrolimus, use of cyclosporine, and trough levels of both), the association of PCSK9 with NODAT remained materially similar (HR 1.39 [95% CI 1.13–1.72]; P = 0.002) after adjustment for age, sex, eGFR, proteinuria, time since transplantation, and use of tacrolimus, use of cyclosporine, and trough levels of both. In an additional Cox regression analysis with PCSK9 divided in the upper versus the two lower tertiles, PCSK9 was again significantly associated with development of NODAT independent of potential confounders (HR 2.37 [95% CI 1.45–3.87]; P = 0.001) after adjustment for age, sex, eGFR, proteinuria, time since transplantation, and use of tacrolimus, use of cyclosporine, and trough levels of both. Secondary analyses in which adjustment for closely related covariates within biological domains were exchanged for adjustment by the variable not included in the primary analyses are shown in Supplementary Table 2. Results of secondary analyses, in which we repeated the Cox regression analyses with inclusion of the variables in biological domains that were not included, did not differ materially from the results of the primary analyses (Supplementary Table 2).

To evaluate the association of PCSK9 with NODAT without taking time to event into account, we performed crude and multivariable logistic regression analyses equivalent to Cox regression models 1 and 3. The crude analysis revealed an odds ratio of 1.34 (95% CI 1.06–1.69; P = 0.02), and the analysis equivalent to model 3 revealed an odds ratio of 1.41 (95% CI 1.10–1.81; P = 0.007).

In further sensitivity analyses, we tested for interaction by use of statins. We found no indication of significant effect modification by use of statins if statin use and a product term of statin use and PCSK9 were added to the crude model (P = 0.52 for interaction). The same was true if the product term was added to multivariable Cox regression model 7 with additional adjustment for age, sex, eGFR, proteinuria, time since transplantation, and use of statins (P = 0.65 for interaction).

During a median follow-up of 10.0 (IQR 9.7–10.4) years, 123 RTRs died. In univariable Cox regression analysis, serum PCSK9 as a continuous variable was not significantly associated with increased risk of mortality (HR 1.12 [95% CI 0.95–1.31]; P = 0.17). After adjustment for relevant covariates, there was also no association of PCSK9 with all-cause mortality (Supplementary Table 3). Of the 123 deceased RTRs, cause of death for 51 (41%) was cardiovascular related. There was no significant association between PCSK9 and cardiovascular mortality (Supplementary Table 4).

During a median follow-up of 9.6 (7.1–10.2) years, 59 RTRs developed graft failure. In univariable Cox regression analysis, serum PCSK9 as a continuous variable was not significantly associated with graft failure (HR 0.77 [95% CI 0.57–1.02]; P = 0.07). After adjustment for relevant covariates, there was also no association of PCSK9 with graft failure (Supplementary Table 5).

In a large cohort of stable RTRs, we show that higher serum PCSK9 levels are associated with development of NODAT. This association remains present independent of relevant confounders and independent of adjustment for use of statins. The data agree with the concept that the PCSK9 pathway plays a pathogenic role in the development of diabetes in this patient category. Furthermore, the results provide a rationale for investigating the extent to which PCSK9 inhibitor treatment would prove beneficial with respect to prevention of NODAT in RTRs, besides its well-established LDL cholesterol–lowering effect.

To date, the association of PCSK9 levels with NODAT has been investigated neither in RTRs nor in other transplanted populations. In renal patients, PCSK9 has been documented to be elevated proportional to the degree of proteinuria (24) but to be unaffected by moderately compromised renal function (25). In the current RTR cohort, PCSK9 was increased in conjunction with (central) obesity, as observed in nonrenal populations (26–28). Thus, although we did not enroll a control group, serum PCSK9 levels appeared to be elevated in a considerable number of RTRs. Many reports have revealed positive correlations of circulating PCSK9 with LDL cholesterol (24,26,28), but such a relationship was not evident in the current study probably because PCSK9 increased the effect of statin treatment (24,25), thereby masking a positive PCSK9-LDL cholesterol relationship. This relationship emphasizes the necessity to adjust for statin use in multivariable analysis regarding the association of PCSK9 with NODAT. In keeping with data in nonrenal populations (24,26,28,29), serum PCSK9 was associated with triglycerides, which we also controlled for in the analysis. Moreover, in keeping with earlier reports (26,30), serum PCSK9 was positively associated with HOMA-IR, which is a known predictor for NODAT (31). Thus, the association of PCSK9 with NODAT as documented here was essentially unaltered, taking into account the use of statins and other covariates, including HOMA-IR. A potential role of statin treatment was also evaluated, but we observed no interaction.

PCSK9 is a main regulator of LDLR availability, and the findings regarding the PCSK9-NODAT association agree with the hypothesis that cholesterol metabolism is linked to diabetes development by involving the LDLR pathway. PCSK9 was positively related to insulin and HOMA-IR, again in keeping with other data (26). The precise mechanisms by which the PCSK9 pathway are involved in diabetes development need to be delineated in more detail in future studies. In comparison, 6–18 months of treatment with the PCSK9 inhibitor alirocumab did not affect transition to new-onset diabetes in populations at high cardiovascular risk, but longer follow-up is required to more conclusively rule out any association (32).

Although evidence is mounting in favor of a beneficial effect of PCSK9 administration on incident CVD (33–35), remarkably few studies have addressed whether circulating PCSK9 predicts cardiovascular outcome. In a Swedish cohort study, PCSK9 was found to independently predict incident CVD during 15 years of follow-up (36). In contrast, PCSK9 was unrelated to newly developed CVD during 17 years of follow-up in the Women’s Health Study (27). PCSK9 was not significantly associated with CVD events or with renal function decline during 3–10 years of follow-up in two German cohorts comprising subjects with moderately impaired kidney function (37). In the current study, PCSK9 did not predict all-cause and cardiovascular mortality. Given the resemblance of transplant vasculopathy, an important cause of graft failure, with atherosclerosis (38), we also determined whether PCSK9 predicts renal outcome. PCSK9 was not related to graft failure in the current study.

This study has several strengths. To our knowledge, it comprises the largest prospectively followed cohort of RTRs in which PCSK9 was measured. Moreover, end point evaluation was complete in all patients despite a considerable follow-up period. Another strength of the study is that we included only stable RTRs who were >1 year posttransplantation, resulting in exclusion of transient posttransplantation hyperglycemia in NODAT diagnosis. Hyperglycemia is extremely common in the early posttransplant period and can occur as a result of rejection therapy, infections, and other critical conditions. Therefore, the formal diagnosis of NODAT in RTRs should only be based on likely maintenance immunosuppression, stable kidney function, and absent acute infections (39).

The study also has several limitations. It was carried out at a single renal transplant center, so the extent to which circulating PCSK9 predicts NODAT development must be replicated in other RTR populations that include sufficient numbers of nonwhite participants. We did not have repeated PCSK9 measurements, and the use of a single measurement of the variable of interest may have given rise to an underestimation of an otherwise existing true effect (40). Another limitation is that we did not perform an oral glucose tolerance test at baseline or during follow-up, which is considered the gold standard test to define NODAT. Thus, we could have underestimated the frequency of NODAT compared with other studies. However, we diagnosed NODAT according to International Expert Panel recommendations that were based on American Diabetes Association criteria (2), which is in line with the common way in which NODAT is diagnosed in transplant centers. Furthermore, as with any observational study, unmeasured or residual confounding may have existed despite the substantial number of potentially confounding factors for which we adjusted. Finally, the observational nature of the study did not allow us to discern whether higher PCSK9 is a cause of NODAT or merely a marker of a higher risk for NODAT.

In conclusion, higher serum PCSK9 is associated with an increased risk of NODAT in RTRs independent of potential confounders. Inhibition of PCSK9 may be a promising target after transplantation to diminish the risk of developing NODAT while having a beneficial effect on hypercholesterolemia.

Acknowledgments. This study was based on the TransplantLines Insulin Resistance and Inflammation (TxL-IRI) cohort study. PCSK9 was measured at Lilly Research Laboratories, Eli Lilly and Company (Indianapolis, IN) for free.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. M.F.E., D.M.Z., S.J.L.B., and R.P.F.D. wrote the manuscript, analyzed data, and performed the statistical analysis. J.H.S. and S.J.L.B. acquired the data and performed the research. C.A.J.M.G., S.J.L.B., and R.P.F.D. provided critical review, advice, and consultation throughout. M.F.E., D.M.Z., J.H.S., C.A.J.M.G., S.J.L.B., and R.P.F.D. read and approved the final manuscript. R.P.F.D. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.